[This post was accidently embargoed due to a bug in our handling of WordPress drafts. It was sent out on the feed in October, but never appeared on the blog page.]

Chris Masse
provided a pointer to a
consent order
between the CFTC and TradeSports
(a.k.a. TEN, the Trade Exchange Network). The
is interesting because in allowing TradeSports to escape punishment by
paying a fine and agreeing not to sell options on commodities
(“without admitting any wrongdoing”), the CFTC is implicitly not
challenging TradeSports’ other markets. This appears to be good news.

The particular markets that the CFTC objected to were markets on the
prices of various commodities (gold, crude oil, and exchange rates)
and speculation on the FOMC’s rate announcements. There’s no mention
of TradeSports’ markets on supreme court appointments, elections
(foreign or domestic), pending legislation (Social Security Reform),
or (!) stock market indexes in several countries. The CFTC could have
declared that by making a market in instruments whose value depends on
the outcome of external events, TradeSports was creating a new
commodity. That would have been sufficient for the CFTC to go after
them. The fact that all these instruments are ignored in the ruling
seems to me at least a signal that the CFTC isn’t intending to try to
stop any of the rest of it. (They can always change their mind
later.) That doesn’t provide any protection from the gambling laws,
of course, but the International Court’s decisions seem to be helping on that front.

The CFTC also points out how cooperative TradeSports has been through
the whole process. Since they are applying to have a subsidiary
covered by the CFTC, this is no surprise.

One of the points in the order is that TradeSports “shall inform
residents of the US of what contracts are unavailable to them for
trade by utilizing a pop-up notice that will appear when such
residents attempt to enter orders on those contracts”. They don’t
seem to have implemented this yet: I was able to enter an order on
the FOMC’s December decision on the funds rate.

This article is the first of a series describing the variety of prediction market institutions in use. I’ll start with the simplest form of prediction markets: a double auction (buyers bidding prices up, sellers bidding down) for a simple asset that represents the outcome of interest. The asset will pay $1 if some outcome comes to pass, and will be worthless if not. This article describes three variations on this basic model.

The calculation of value for the buyers is simple: if the asset’s expected value is higher than the best price asked by a seller, then they should buy. Buyers who are not worried that new information will change the value before they can withdraw their offers should be willing to leave standing orders that prospective sellers can accept.

The calculation for the sellers is more subtle. Those who believe the asset’s expected value is lower than the best current price offered by a buyer should sell short, accepting the current bid price in exchange for a liability that they may have to pay back later. Fortunately, the liability is limited; the most they will have to repay is $1 for each share. The market operator will usually have a mechanism for managing their cash to ensure they will be able to pay back the liability if the outcome goes against them. Until the maturity date of the asset, they have the proceeds to invest as they think prudent.

That’s the starting point for Prediction Markets. Let’s call it the “short selling model”, since that’s the most salient feature. It’s sufficient to model markets that provide predictions (by publishing the market consensus of the estimated probability that an event will occur), encourage research, reward insight, and allow insurance and hedging of exposure to various risks.

The first variation defines two complementary assets in place of the original one. The second asset pays $1 when the first is judged at 0, and pays nothing when the first is decided to be worth $1. This variation was first proposed by Robin Hanson in 1988. This version makes the process symmetric between the seller and the buyer. They both spend money to buy assets that may pay off at some date in the future.

The short selling required in the first model can be a psychological barrier to some traders, who may not see that it differs from short selling in the stock market. In the stock market, the assets have unlimited growth potential (consider Google), which means that the liability can also grow without bound. Prediction market assets have a known maximum payout, which means the maximum liability is limited as well. While the risks and reward are the same in both versions, people’s unconscious aversion to short selling may make them more hesitant to sell than to buy.

The variant (call it “buying complementary assets”) has both sides buying assets, so it sidesteps the psychological barrier. Another variation (called “buying a basket of assets” or just “buying a basket”) also uses complementary assets to eliminate short selling, but selling doesn’t look the same as buying. In this version, the market is always willing to buy complementary pairs from traders (or sell them) for $1. Since one or the other of the assets will pay $1, this is an equal trade. The difference in this model is that a trader who want to bet against a position does so by buying a pair of assets, and selling the unwanted position. Buying the pair and selling one asset has the same effect on a trader’s holdings as buying the complementary asset. Since the difference can also be hidden in the user interface, the only advantage I know of for the “basket” model is that in a play money market, users can start out with balanced baskets of assets to encourage trading.

After a trade, the buyers in the three variations all have the same portfolios: if the price was p, their cash is reduced by p, and they have one additional positive asset. The sellers in the complementary asset and basket models are in a similar position: their cash is reduced by (1 – p), and they have an additional positive asset (though of the complementary position). Sellers in the short selling model is the only one that’s different: their cash is increased by (1 – p), and they are “short” one unit of the asset. If the asset ends up in the money, they owe that money to the market. Of course, after taking any of these positions, traders can sell or buy the same asset in order to close their position. They can do this to lock in gains, or to cut losses, but either way, they’re contributing to the price-setting activity of the market.

TradeSports (aka InTrade, TEN) uses the short selling model. Since they come to Prediction Markets from the betting world, they use the short selling as a way to keep more money in their customers’ accounts. When you sell short, your balance increases. The software reserves margin funds as assurance that traders will be able to cover their short positions, but it gives a generous allowance for claims that won’t settle for a while, and for offsetting positions in each trader’s account. Since traders are often more willing to bet in favor of a proposition than against, this added incentive to sell is probably a good trade-off for TradeSports.

The Iowa Electronic Markets (IEM) and NewsFutures follow the “buy a basket” model. On IEM, there is an explicit step (and a separate screen) for buying baskets of assets. Traders who don’t master this step can only buy assets, but can still bet in either direction, since there are separate markets in the different assets. I’ll talk later about why separating the markets is a mistake.

On NewsFutures, the explicit step of buying the basket has been hidden in the common case where there are only two outcomes; it’s only visible in multi-position claims. NewsFutures doesn’t currently have any multi-position claims, so I can’t verify this. (I’ll talk about markets with multiple outcomes in a later posting.) The Foresight Exchange follows the simpler “buying complementary assets” model.

Market operators who depend on trading for their profits want to encourage more trading. One of the obstacles is that traders are relatively unwilling to leave orders on the books (some estimates are that 80 percent of traders seldom or never leave standing orders.) If the market operator can encourage traders to leave more standing orders, there will be narrower spreads, and more trading opportunities. IEM is non-profit and NewsFutures and Foresight Exchange use only play-money, so they don’t take extra steps to encourage book orders. TradeSports uses its fee structure to encourage book orders. Only the person who accepts an offer out of the trading book pays a transaction fee. The trader who left the standing order trades for free.

From these basic market foundations, variations have been developed that support asking complex and interdependent questions, that support questions the basic mechanism can’t answer, and that predict well even when there are few traders. I plan to talk about all the variations in later posts.

Salon has a complimentary (but not free) article on InTrade’s Bird Flu markets, but they seem to overbill the minor points and underplay the important questions and effects. They’re interested in whether it’s immoral to bet on events that will turn out to be a tragedy for someone. “Is this market offering the chance to win blood money?”, and “[Do traders] worry about profiting from tragedies?”, but they can’t seem to get a rise out of either the people they interview or anyone on InTrade’s message boards. They quote someone on the message board who points out that moral hazard is the relevant ethical risk, and that doesn’t seem to be present for acts of nature like bird flu.

The more interesting question is whether the bird flu market can tease out latent information and make it accessible to market watchers. Salon points to markets that have been successful (and ones that have overclaimed success), and seems to assume this means we can rely on the bird flu market’s predictions. This market seems different enough from all the others that have been tried that I’m waiting for more data before relying on the results.

On the other hand, Salon’s tag line “If the price of the bird flu spikes, it’s time to prepare for the worst…” has a little to recommend it. At least if the price spikes, I’ll expect someone to investigate and tell us why.

I took pictures at the Prediction Market summit a week ago. I managed to capture all the speakers other than myself; I inexplicably neglected to get a picture of John Maloney, who organized and ran the meeting. I also took snapshots of the audience. Rohit also took pictures.

I will be speaking at the next Summit, which will be in New York, and will have more of an East Coast flavor. Robin Hanson will be there, and the Keynote will be by James Surowiecki, author of “The Wisdom of Crowds”. Expect other speakers to be announced shortly.

I added a PDF version of the slides from my talk to the PowerPoint I posted before.

The economist has a
short article
on prediction markets, suggesting that they should
be used inside companies a lot more, and that a prime area where they
have been underutilised is in predicting technology trends. They quote
or cite most of the big names in the field.

The main problem with trying to use Prediction Markets to foresee
trends is that you have to ask the right question. Yahoo’s

Tech Buzz
game at least has the ability to list several competitors in a field
and let the market predict how they will each do. But so far, I doubt
that is giving more insight than just watching Yahoo’s search ratings
would provide. And it doesn’t help at all when you’re looking for the
next innovation to take off. Before the introduction of the iPod, the
market couldn’t have told you that

Portable Media Devices
was a good technology to develop.

The closest example that I know of is

Ely Dahan’s
work on Prediction
Markets as a replacement for focus groups. This approach will help if
you’re deploying a product, and want to know what product features
will be valuable, but I don’t think Dely has shown that it can be
successful when you are working on a breakthrough technology; all the
product combinations he’s talked about so far were already widespread
in the market when his groups evaluated them. If the market
participants haven’t seen how the new product would fit in their life,
they may not be able to evaluate it properly.

I have posted my slides from my talk about the Prediction Market summit, so I now feel free to bug others to post theirs. The Prediction Market Summit last week was quite a success. We had quite a variety of speakers. I led off the day by summarizing Zocalo’s current status, and showed a replay of an experimental session. I then pitched the idea that Prediction market operators should make their sites more searchable. This would raise the visibility of prediction markets generally, as well as their specific markets.

I spent the rest of my time explaining a way that Prediction markets based on claims with multiple exclusive outcomes should be presenting better prices to traders. These markets maintain a separate double auction for each outcome, segmenting the available order volume into non-interacting submarkets. Arbitrageurs are not a substitute for the market in offering these trading opportunities, since riskless arbitrage can’t take advantage of interest that never appears in the order book. After a few hallway conversations I think most people understood the argument.

Bo Cowgill talked about Google’s internal markets. They wrote their own software from scratch, following the IEM model. Traders buy a basket of claims in order to sell a claim. Usability and politics were the biggest issues for them. The company contributes money in an account for each Google employee, which they can then trade. Each participant’s account is cashed out at the end of each quarter. They get as many lottery tickets as they had cash from liquidating claims. (i.e. money you don’t invest doesn’t earn anything in the lottery.) Prizes are then awarded to lottery winners; this eliminates the incentive (common in play money markets with prizes for top performers) to take extreme chances in order to boost your odds of being the single top performer. A legal issue they had to contend with in selecting claims to bet on was that information on some outcomes is controlled by the SEC. If employees find out what is happening in some areas, they might become subject to SEC rules for insiders, and their stock trading restricted. The operators of the market chose subjects where that likelihood seemed small.

Bernardo Huberman presented work at HP on techniques for using games to elicit trader’s risk preferences and subject knowledge, and then use nonlinear functions to aggregate their predictions. This approach can address problems due to illiquidity when there are few traders, and prevent manipulation. Bernardo reported that HP is starting to use some internal markets to predict price and availability of various PC components on which the company depends.

Emile Servan-Schreiber reported that the prediction market business climate is improving. NewsFutures has been getting better responses as they sell the concept within businesses. He listed Dentsu, Lilly, Mars, Abbot, Yahoo, and the World Economic Forum as customers who are currently using their software.

Mike Knesevitch talked about TEN’s (Trade Exchange Network, also TradeSports, and InTrade) markets. The fact that they trade in real money allows investors to hedge. Mike said their sports markets are extremely efficient because there are a number of arbitrageurs who know the statistical relations between various outcome, and take advantage whenever prices exceed certain bounds. Since they don’t take a position in any of the trades, Mike said they aren’t subject to the 1961 Wire act, which regulates interstate gambling in the US. It also means they couldn’t get a bookie’s license in the UK if they wanted one. TEN has approval from the CFTC to operate as an exempt Board of Trade. eBOTs can only support trading between certain large firms and qualified investors. My understanding of what Mike said is that TEN intends to operate a separate exchange for these large players, allowing them to hedge positions by trading in prediction markets. One area TEN intends to grow in is weather contracts. Mike said TEN, as one of their criteria for approving a new claim, looks for natural trading partners who would take opposite positions. He pointed out that Ski resorts (whose business falls in a light snow year) have exposure that is opposite to that of big cold weather cities (whose snow removal expenses grow in a heavy snow year.)

Russell Anderson talked about HedgeStreet’s markets. One factoid I picked up is that HedgeStreet gets their piece by charging a percentage of the winning side of contracts. TEN charges a commission to traders who buy at the market price. Orders entered into the book are free.

Eric Zitzewitz presented his paper with Justin Wolfers on Interpreting Prediction Market Prices as Probabilities. This paper is a response to Charles Manski’s earlier paper arguing that prices are more likely to constitute a weighted average of beliefs. Eric argued that a more general model of trader’s preferences and budget leads to a model in which prices are very close to probabilities, and that traders have practical incentives to move the prices closer to probabilities. The paper shows that their model also predicts price curves that match the results seen in actual markets more closely than Manski’s do. One good line that Eric used in explaining the uses for Prediction Markets is that they allow us to study the likelihood of events even when the events never actually take place.

Todd Proebsting has been proselytizing for Prediction Markets inside Microsoft for a couple of years. He set up markets that continue to be used for internal decision making, though not in any formal routine process. The first market they ran quickly showed that a project was in serious schedule trouble, even though the management team had been reassured (by the same employees who participated in the market) that the project was on track. Unfortunately for the people who predicted the schedule problem, the project manager changed the milestones in response to the dire predictions, eliminating the value of the investments based on the completion date. Since then, Todd has been more careful about how questions are phrased. Microsoft’s markets uses a market maker and doesn’t support limit orders because their initial informal survey showed that people were hesitant to leave book orders. Todd recently saw Edward Tufte’s presentation on the evils of PowerPoint, so he spoke without slides which might have been illegal if he tried to do it at Microsoft.

Dave Pennock gave the audience a choice among a few short presentations he had prepared. The audience opted first for his presentation on Search Futures and the Tech Buzz game. The Tech Buzz forecasts got better as deadlines drew nearer. Some traders were able to figure out that the prices in the Dynamic Parimutuel market should be proportional to the square root of the underlying index value (search terms on Yahoo); the best traders pushed prices closer to that level. The second section of the talk covered the paper Pennock wrote with Servan-Schreiber, Wolfers, and Galebach. They analyzed data from real money and play money prediction markets for NFL football and showed that the prices on both were much better predictions than those of most bettors. (NewsFutures and TradeSports came in 6th and 8th out of about 2000 contestants.)

The day ended with a panel that I moderated. I didn’t get a chance to take notes, so all I can say was that it was congenial, and the panelists addressed all the questions.

The format was generous, with plenty of time between sessions for people to talk. The location at UCSF’s Mission Bay campus worked quite well.

The Prediction Market Summit
that CommerceNet sponsored in San Francisco last week was quite a
success. There were 9 speakers, including four that I hadn’t heard
before. I’ll have more to say about the talks and I’d like to post
some of the pictures I took, but I’m going to have to dole it out over
several days so I can get other work done. The first thing I want to
take care of is posting my slides. I dropped out of PowerPoint
twice, so I want to make a version of those exhibits that I can
present with the slides. The first time I
dropped out of PowerPoint was to show a replay of a session of an
experiment the folks at GMU ran using Zocalo in early November.

In the experiment, forecasters watch the market interaction, and try
to decide whether the value of the tickets being traded is 0 or 100.
4 pure traders and 4 manipulators are active in the market. Each
starts out with two tickets and 200 units of currency. Both the
traders and the manipulators have each been given a hint (the hints
are true with 2/3 chance) as to the actual value of the tickets. The
manipulators have an additional incentive to make the forecasters
believe that the asset has one value or another. Sometimes the
incentive is consistent with the hint, sometimes not.

You can see the replay here.
The javascript that drives it only works in FireFox, so you won’t be
able to view it with other browsers. (I will look into getting a quicktime or other movie made.) The image above displays the
contents of the order book (green and orange dots) at the time of each
transaction (black dots). The UI that the experimental subjects see only
shows the current contents of the order book. Notice that the order book in the center of the screen updates along with the colored dots in the strip chart. The “immediate order” buttons are updated as orders are entered and filled.

I created the depiction
above in hopes that the extra context it gives would be useful for
figuring out how much interest there is on each side of the market.

Chris Masse sent me a pointer to Marco Ottaviani‘s draft announcement for a Mini-Conference on Information and Prediction Markets to be held December 19 in London. That’s so little notice, and so close to Christmas that I’m not seriously contemplating attending, but the list of speakers is another good one. If London isn’t a long trip for you, think about this one.

  • Paul Tetlock: Designing Information Markets for Decision Making
  • Robin Hanson: Comparing Ways to Encourage Information Market Participation
  • Leighton Vaughan-Williams: Efficiency in Exchange Betting Markets
  • Olivier Gergaud: Strategic Forecasting in Rank-Order Tournaments
  • Frederic Koessler: Individual Behavior and Beliefs in Experimental Parimutuel Betting Markets
  • Justin Wolfers: Interpreting Prediction Market Prices as Probabilities
  • Peter Norman Sorensen: Aggregation of Information and Beliefs in Prediction Markets
  • Erik Snowberg: Partisan Impacts on the Stockmarket
  • Koleman Strumpf: Manipulating Political Stock Markets
  • Eric Zitzewitz: Full Distribution Event Studies

It’s nice to see the number of conferences and workshops in this area growing.

The Prediction
Markets summit
that will be held in San Francisco on December 2nd
is shaping up to be quite an event. There will be speakers from
Google, HP, NewsFutures, Trade Exchange Network, HedgeStreet,
Stanford, Microsoft, and Yahoo. I’ll start things off by telling what
I’ve been able to do with Zocalo so far, and then challenge the other
attendees on a few points of Prediction Market design. Bernardo
Huberman of HP will give the keynote. Google and Microsoft will talk
about their experiences with markets inside the corporation.
Representatives of NewsFutures, InTrade, HedgeStreet, and Yahoo will
talk about how they are faring in the market. Eric Zitzewitz of
Stanford will present a
responding to
charge that prices in prediction markets don’t reflect

The day will end with a panel discusssion, which I will facilitate.
I’ll ask about Tom Bell’s

draft legislation
, about the prospects for
increasing liquidity on HedgeStreet, and I’ll try to get InTrade to
say what they plan for their proposed

US subsidiary
if they haven’t talked about that earlier in the day.

This will be the first public opportunity that I’m aware of to see
presentations on

Google’s markets
, HedgeStreet, or InTrade. There is
still space available. If you are at all interested in

, you should register now.

The whole event is billed as having a conversational, interactive
style, so if you want to find out more about these ventures, this will
be the place. Future events in the series are planned for the east
coast, Europe, and Asia, but they’ll have very different speaker

Tom Bell
posted a short draft (with variations) of a bill legalizing Prediction
Markets. The
bill’s focus is explicitly on markets for science, technology, and
public policy, so it would benefit the
Simon Exchange and possibly
the Washington Stock
, but not other, general purpose, markets. The strongest
version’s provisions are really simple: no federal, state, or local
government may enact or enforce laws to regulate prediction exchanges,
and standard commercial law continues to hold regarding the contracts.
(The limitation to claims on science, technology, and public policy
is in the definitions.) The weakest version only forbids enforcement
against prediction markets of laws concerning gambling, commodities
futures, securities, or insurance. I’ve simplified the legalese in
order to get it to a couple of lines. Tom’s version is only a little
longer in order to have all the necessary legalese.

These would be very nice rules if they were law, but the more
interesting question is what it would take to get them enacted.
Congress isn’t in the business of enacting legislation merely because
it would make us all freer; there has to be a substantial lobby, or
evidence of plenty of interest to make it happen. In order to get the
support of the relevant industry (Trade
Exchange Network
, ProTrade),
the terms would have to be opened up to include more than just science
and technology. So a big question is whether enough support could be
lined up with a version limited to science and technology to get a
bill passed (it should be easier because the subjects are respectable,
and can be pitched as “not like gambling”), or whether having
additional companies backing it would help more. I wouldn’t mind
letting the sports markets join the party, as long as it makes it easier for a bill to pass. If no bill is going to
pass, it doesn’t make a difference anyway, and we’re left looking for
legal strategies that will allow public policy markets to avoid a suit
or survive one if it happens.